Crowd Post-Correction of HTR Output in a Pedagogical Context: The Case of the Paris Bible Project 'Correct-a-thon'

Abstract

Crowd transcription has caught on in recent years and has been bolstered by the design of different interfaces to match the objectives of content co-creation. In January 2023, our own transcription event took place in Besançon, France. It grew out of the Paris Bible Project which aims to understand the production and diffusion of the mass-produced biblical manuscripts in medieval Europe using digital methods. Inspired by two events organized for medievalists to transcribe multiple witnesses of the same medieval textual tradition, ours was a week-long event developed in collaboration with master’s students at the Université de Franche-Comté aimed to model forms of collaborative research-based inquiry in collaboration with HTR. In our paper, we argue that since there is a large variety of expert and semi-expert communities interested in the future of the "transcribe-a-thon," we need to design such events both for these communities as well as for the potential reuse of their data. We also argue that, while such community events are often sponsored by research-led projects, they should not only be held to improve the quality of data, but they should also be framed as opportunities to learn elements of data literacy, particularly when the "academic crowd" is made up of students.